BackgroundBrain tumor related epilepsy (BTE) is a major co-morbidity related to the management of patients with brain cancer. Despite published practice guidelines recommending against anti-epileptic drug (AED) utilization in patients with gliomas, there is heterogeneity in prescription practices of AEDs in these patients. In an attempt to impact BTE management, we statistically analyzed clinically relevant attributes (sex, age, tumor size, tumor growth kinetics, and tumor location) pertaining to seizure at presentation and used them to build a computational machine learning model to predict the probability of a seizure (at presentation). Methods From our clinical data repository, we identified 223 patients (females, n=86; males, n=137) with pathologicallydetermined glioma and known seizure status at clinical presentation. Non-parametric and Fisher's Exact tests were used to identify statistical differences in clinical characteristics. We utilized a random forest machine learning method for generating our predictive models by entire cohort and separated by male and female. Findings Patients were divided into those that presented with seizure (SP, n=96, 43%; F, n= 28; M, n= 68) and those that presented without seizure (nSP, n=127, 57%, F n=58, M n=69). Females presented with seizures significantly less often than males (x 2 =6·28, p=0·01). SP patients had significantly smaller T1Gd radius compared to nSP (SP 11·30mm, nSP 18.66mm, p<0·0001). Tumor size and patient age were significant negative predictors for SP; patients with larger tumors, older age and less tumor diffusivity (p/D) were at lower risk for SP. Interpretation Despite heterogeneity across our patient cohort, there is strong evidence of a role for patient sex, tumor size, tumor invasion, and patient age in predicting the incidence of seizures at diagnosis. Future studies, with prospectively detailed data collection, may provide clearer insights into the incidence of seizures through a patient's treatment course.